Passive Microwave Remote Sensing of the Atmosphere and Ocean
Total Page:16
File Type:pdf, Size:1020Kb
Proc. Indian Acad. Sci. (Engg. Sci.), Vol. 6, Pt. 3, September 1983, pp. 233-254. Printed in India. Passive microwave remote sensing of the atmosphere and ocean T A HARIHARAN and P C PANDEY Space Applications Centre, Ahmedabad 380053, India Abstract. The microwaveremote sensing experimentsconducted during the last decade using airborne and spaceborne sensors are now evolving as operational spacecraft observation systems. Microwave sounding unit (MStJ) onboard TmOS-~qregularly provides global atmos- pheric temperature profile required for numerical weather prediction. The monitoring of composition profile and geophysical parameters from space platform is now a reality and holds a great promisefor future meteorologicaland oceanographicresearch. This reviewpaper summarises the recent advances and future opportunities in passive microwaveradiometry. Indian microwave remote sensing programme and achievements to date are also described. Keywords. Passive microwave radiometry; retrieval technique; atmospheric sounding; ter- restrial sounding; satellite microwaveradiometer; Bhaskara satellite 1. Introduction During the past decade, the achievements from the NEMS (Nimbus E microwave spectrometer) and SEAMS (scanning microwave spectrometer) have led to the develop- ment of the first operational microwave spectrometer flown on the TmOS-rq series of satellites. These satellites carried a four-channel microwave sounding unit (Msu) for sounding the atmosphere. Also, a seven-channel microwave temperature sounder was flown on a Block D meteorological satellite developed by defence meteorological satellite program (DMSr,). The launch of SEASAr and Nimbus-7 satellites and the successful operation of the various instruments onboard the spacecraft was a great technological advancement. It was a proof of the concept mission and has added a new dimension in our capability of monitoring the ocean from space. Global measurements of sea surface temperature, wind speed, precipitable water and sea-ice coverage with significant accuracies are a reality now and holds a great promise for future operational applications. The principal advantage of microwaves is their relative ability to penetrate clouds, haze, precipitation and surface material such as snow and ice. Besides, other properties unique to microwaves can also be exploited. Tomiyasu (1974), Ulaby (1976), Swift (1980), Staeli n ( 1981) and more recently Njoku (1982) have reviewed the advances in the field of microwave remote sensing of the earth's surface and its atmosphere. The present paper reviews the progress to date in satellite microwave radiometry with emphasis on the current important results obtained from SEASAT SMMR (scanning multichannel microwave radiometer) and SAMm (satellite microwave radiometer) onboard the second Indian earth observation satellite Bhaskara. 2. Basic principles of passive microwave radiometry Thermal radiation emitted by geophysical sources is expressed in terms of brightness temperature TR(v), in units of degrees Kelvin, where T~(v) is the temperature of a black- 233 234 T A Hariharan and P C Pandey body radiator that produces equivalent power at frequency v. The values ofT8 (v) range between 60 and 305 K for oceans and tropical forests respectively. The brightness temperature observed from satellite-borne microwave radiometer, is given by radiative transfer theory described in detail by Chandrasekhar (1960) oo h' TB(v) = [Toe + (1- e) f T (h)at(h)exp ( - f a(h') dh' ) dh ] exp ( - zo ) o o oo oo +f T(h)ot(h)expl-f ct(h)dh]dh, (1) o h where T(h) and at(h) are the atmospheric temperature (K) and absorption coefficients (m- 1) respectively as a function of height h. To is the surface temperature and e is surface emissivity. oo ~o = f a(h')dh' = total atmospheric opacity. (2) 0 In case of scattering from rough surfaces or from atmospheric constituents like hydrometeors, the above equation of radiative transfer is modified (Tsang & Kong 1976). Matrix equation of radiative transfer is required for considering the effect of mesospheric oxygen which is influenced by magnetic field (Lenoir 1967). This will be useful for studying atmospheric constituents using microwave limb sounder. Equation (1) assumes an approximately specular reflection at the surface. The equation was computed by dividing the atmosphere into concentric layers, each with a specified temperature and absorption coefficient and using a summation to approxi- mate the integrals. Cloud layers can be inserted at any altitude by modifying the absorption coefficients to include an appropriate amount of liquid water. In addition, the relative humidity of the air was assumed to be 100% in the presence of clouds. The principal absorbing constituents of electromagnetic radiation are 02, H20, hydrometers, and many trace gases including 03, CO, H20, HNO3, different oxides of nitrogen, O, OH, C10, H202 and others. An excellent treatment of the absorption due to various atmospheric gases and trace constituents is described by Waters (1976). The accuracy requirements of various absorption formalisms are adequate for general remote sensing applications but there is still some debate on the role of water vapour dimers and its effect on remote sensing, which may be important if relative humidity is high. This and some other issues like line shape and in some cases line strength require further research. 3. Retrieval techniq.,~e Geophysical parameter retrieval techniques using multiwavelength microwave radiometric data are statistical (Waters et a11975; Grody 1976; Wilheit & Chang 1980; Hofer & Njoku !981; Pandey & Kakar 1983), nonlinear iterative (Wentz 1983) and fourier transform technique (Rosenkranz 1978). The most generally used procedure is Passive microwave remote sensin 9 235 statistical, which was first used by Waters (1975) for retrieving atmospheric tempera- ture profile from measurements near 60 GHz 02 band. The principle behind this method is to find the linear predictor D = Dy, (3) by minimising E { (x -s (x- s ~ is the estimate of parameter vector and y is the measurement vector. This is of course classical problem of multiple linear regression, with elements of D as regression coefficients. The matrix D is given by l) = e{x. yr} (e{yyr})- ~. (4) Both the expected values in the above expressions are covariance matrices. Instead of evaluating D as given in (4), sometimes it is more convenient to subtract mean ofx and y to obtain the covariance matrices. If we have a large sample of measured Y with our radiometer and also have independent direct measurements from radiosonde, ships or rock~tsonde, then we can estimate 1) entirely from experiment. Otherwise radiative transfer modelling with known physics and the a priori statistics is used to estimate I). Incorporation of a few nonlinear terms in the above retrieval can, sometime, improve the accuracy (Sharma et al 1981; Rosenkranz et al 1972). In the temperature profile retrieval, an improvement of 10 ~ 30 % is obtained using Kalman filtering technique (Ledsham & Staelin 1978). However this technique has not been applied to retrieve geophysical parameters from multiwavelength measurements and holds promise for future research in this area. For nonlinear problems, analytical technique developed by Chahine (1977) can also be used, specially for profiling of water vapour, temperature and atmospheric composition. 4. Geophysical applications 4.1 Atmospheric sounding Waters et al (1975) reported the first retrieval of temperature profile from NEMS onboard Nimbus-5 satellite. This instrument measured atmospheric radiation near 60 GHz oxygen band. The mixing ratio of oxygen in the atmosphere is quite uniform and time-invariant; thus measurement of atmospheric emission at ,,, 60 GHz is proportional to atmospheric temperature at altitude levels defined by temperature weighting functions (Meeks & Lilley 1963). The three frequencies used for temperature retrievals were 53-65, 54.9 and 58-8 GHz. In addition, two more channels 22-235 and 31.4 GHz were also used for precipitable water retrieval. The NEMS instruments for temperature retrieval were further improved and were followed by similar instruments with more number of channels on TIROS series and the block 5D satellites. A typical weighting function for block 5D temperature sounder is shown in figure 1. The NEMS results have demonstrated the potential of passive microwave techniques for remote sensing of atmospheric temperatures from earth orbiting satellites. An RMS accuracy of ~ 2~ has been reported by Waters (1975) based on comparison of NEMS- 236 T A Hariharan and P C Pandey 40 32 GHz 69.6 GHz | "o 58.825 GHz ..J .••58./.,~/- 54.9 GHz :\ \ 35 GH, - \~,,~,/-53.2 GHz 0.02 0.06 0.10 weighting function ( km -1) Figure 1. Atmospheric weighting function for a downward looking radiometer with an assumed calm sea background for SsM/T on-board block 5D satellite. derived temperature profile with ground truth data obtained from the National Meteorological Centre (NMC) operational analysis and radiosondes. Figure 2 shows NEMS temperature retrieval and RMS variation. The effect of clouds on temperature retrieval was also investigated by Staelin et al (1975) and it was found that clouds affect less than 0"5% of the sounding; the sounding affected most is centred around inter-tropical convergence zone 0TCZ). They also observed a high correlation between 53.65 GHz brightness temperature and NEMS inferred liquid water. Figure 3 shows the result. Further, improvements in temperature profile retrieval are expected with ap- proximately